less than 3 bw auths available: self-measurement (with 10k cap in effect)
Hi, if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available. If you run a fast relay it is capped to 10k cw. This affects currently the 857 fastest relays. regards, nusenu -- https://mastodon.social/@nusenu twitter: @nusenu_
On Tue, Feb 27, 2018 at 06:47:00PM +0000, nusenu wrote:
Hi,
if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available.
If you run a fast relay it is capped to 10k cw.
This affects currently the 857 fastest relays.
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them. moria1 is one that had problems, so I'm hoping to have that resolved shortly. The extrafast relays will get a bit of a respite for now, and I guess the slower relays will have some excitement until we can get things back to normal. Thanks, --Roger
Hi, Roger Dingledine:
On Tue, Feb 27, 2018 at 06:47:00PM +0000, nusenu wrote:
if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available.
If you run a fast relay it is capped to 10k cw.
This affects currently the 857 fastest relays.
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them.
Perhaps it makes sense to do a call and add some more bandwidth authority relays during the upcoming meeting in Rome similar to the Montreal meeting. Cheers, ~Vasilis -- Fingerprint: 8FD5 CF5F 39FC 03EB B382 7470 5FBF 70B1 D126 0162 Pubkey: https://pgp.mit.edu/pks/lookup?op=get&search=0x5FBF70B1D1260162
On 2 Mar 2018, at 12:08, Vasilis wrote:
Hi,
Roger Dingledine:
On Tue, Feb 27, 2018 at 06:47:00PM +0000, nusenu wrote:
if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available.
If you run a fast relay it is capped to 10k cw.
This affects currently the 857 fastest relays.
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them.
Perhaps it makes sense to do a call and add some more bandwidth authority relays during the upcoming meeting in Rome similar to the Montreal meeting.
Would the following documents still be valid (They themselves state they might be outdated)? https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthority https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthorityMeasurem... Also what bandwidth should an bwauth have available itself? I can see if I can support by running one, although it will be EU based. Stijn
On 3 Mar 2018, at 02:15, Stijn Jonker <sjcjonker@sjc.nl> wrote:
On 2 Mar 2018, at 12:08, Vasilis wrote:
Hi,
Roger Dingledine:
On Tue, Feb 27, 2018 at 06:47:00PM +0000, nusenu wrote:
if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available.
If you run a fast relay it is capped to 10k cw.
This affects currently the 857 fastest relays.
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them.
Perhaps it makes sense to do a call and add some more bandwidth authority relays during the upcoming meeting in Rome similar to the Montreal meeting. Would the following documents still be valid (They themselves state they might be outdated)? https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthority https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthorityMeasurem...
Also what bandwidth should an bwauth have available itself?
I can see if I can support by running one, although it will be EU based.
You need a directory authority to vote on your bandwidth authority's output. Bandwidth authorities measure relay capacity. Then they send their results to a directory authority, and the directory authority puts the results in its vote. The directory authority votes change the consensus weights of relays. If your bandwidth authority isn't used for voting or testing, it's just wasting bandwidth. If you want to test and contribute code to a new bandwidth authority implementation, I'd recommend: https://github.com/TheTorProject/bwscanner But you'll need to change the default bandwidth server config, due to the tor project DDoS. T
Hi Teor & Others, Thanks for your response, On 2 Mar 2018, at 23:26, teor wrote:
On 3 Mar 2018, at 02:15, Stijn Jonker <sjcjonker@sjc.nl> wrote:
On 2 Mar 2018, at 12:08, Vasilis wrote:
Hi,
Roger Dingledine:
On Tue, Feb 27, 2018 at 06:47:00PM +0000, nusenu wrote:
if your relays behave strangely in terms of bandwidth seen, than this might be due to the fact that there are less than 3 bw auth votes available.
If you run a fast relay it is capped to 10k cw.
This affects currently the 857 fastest relays.
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them.
Perhaps it makes sense to do a call and add some more bandwidth authority relays during the upcoming meeting in Rome similar to the Montreal meeting. Would the following documents still be valid (They themselves state they might be outdated)? https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthority https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthorityMeasurem...
Also what bandwidth should an bwauth have available itself?
I can see if I can support by running one, although it will be EU based.
You need a directory authority to vote on your bandwidth authority's output.
Do you know what is the best way to get these vote(s), i.e. who to approach; as these kind of things are still a mystery with the Tor project for me. From a personal believe happy to assist, have reasonable spare CPU/Mem and Bandwidth available. So that should't be an huge issue. I know a thing or two about running systems so to say..
Bandwidth authorities measure relay capacity. Then they send their results to a directory authority, and the directory authority puts the results in its vote. The directory authority votes change the consensus weights of relays.
If your bandwidth authority isn't used for voting or testing, it's just wasting bandwidth.
If you want to test and contribute code to a new bandwidth authority implementation, I'd recommend:
Thanks I found the TorFlow repo; that feels a bit hacked, but if either do the job and the above Q can be answered then happy to (try and) set it up.
But you'll need to change the default bandwidth server config, due to the tor project DDoS.
I assume that can be shared in a more private setting then. Stijn
On 5 Mar 2018, at 00:35, Stijn Jonker <sjcjonker@sjc.nl> wrote:
Perhaps it makes sense to do a call and add some more bandwidth authority relays during the upcoming meeting in Rome similar to the Montreal meeting. Would the following documents still be valid (They themselves state they might be outdated)? https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthority https://trac.torproject.org/projects/tor/wiki/doc/BandwidthAuthorityMeasurem...
Also what bandwidth should an bwauth have available itself?
I can see if I can support by running one, although it will be EU based. You need a directory authority to vote on your bandwidth authority's output.
Do you know what is the best way to get these vote(s), i.e. who to approach; as these kind of things are still a mystery with the Tor project for me. From a personal believe happy to assist, have reasonable spare CPU/Mem and Bandwidth available. So that should't be an huge issue. I know a thing or two about running systems so to say..
It's a big ask. You need a directory authority operator to trust you enough to take arbitrary, unvalidated output from a process you run, and feed it to their directory authority. Try to meet some directory authority operators, if you don't know any. Get involved in running relays and other Tor volunteer work. But it might not happen.
Bandwidth authorities measure relay capacity. Then they send their results to a directory authority, and the directory authority puts the results in its vote. The directory authority votes change the consensus weights of relays.
If your bandwidth authority isn't used for voting or testing, it's just wasting bandwidth.
If you want to test and contribute code to a new bandwidth authority implementation, I'd recommend:
https://github.com/TheTorProject/bwscanner
Thanks I found the TorFlow repo; that feels a bit hacked, but if either do the job and the above Q can be answered then happy to (try and) set it up.
We need a bandwidth authority implementation that works better than TorFlow. One thing you can do to help is run bwscanner, and feed its output to a test directory authority. You might find chutney useful for setting up a test network: https://gitweb.torproject.org/chutney.git
But you'll need to change the default bandwidth server config, due to the tor project DDoS.
I assume that can be shared in a more private setting then.
You can set up your own bandwidth server: https://gitweb.torproject.org/torflow.git/tree/NetworkScanners/BwAuthority/R... T
On Tue, Feb 27, 2018 at 02:02:19PM -0500, Roger Dingledine wrote:
Yep! We had 4 running, but 2 of them had problems, and we need 3 for the authorities to want to use the values from them.
moria1 is one that had problems, so I'm hoping to have that resolved shortly.
And all four of them are now back and working. Thanks for your patience everyone. Turns out the issue was that the default bwauth backend (the server that serves the bandwidth files) went offline during our efforts to shuffle things around so www.torproject.org could survive this week's 15-20gbps ddos attack on our website. Faravahar and moria1 were still using the default bwauth backend, but we've moved to a different one and things are looking fine again. Never a dull moment, --Roger
On 03/02/2018 01:17 PM, Roger Dingledine wrote:
Turns out the issue was that the default bwauth backend (the server that serves the bandwidth files) went offline during our efforts to shuffle things around so www.torproject.org could survive this week's 15-20gbps ddos attack on our website.
Faravahar and moria1 were still using the default bwauth backend, but we've moved to a different one and things are looking fine again.
Never a dull moment, --Roger
Thanks for all of your hard work; Tor has good back office. I've been off-line for a while myself because I've been trying to upgrade equipment (& Tor itself -Error 2...?!?), and I'd rather not run relays unless I'm running them according to Hoyle, i.e. up-to-date and secure.
participants (6)
-
Kenneth Freeman -
nusenu -
Roger Dingledine -
Stijn Jonker -
teor -
Vasilis